US6397303B1 - Data processing system, cache, and method of cache management including an O state for memory-consistent cache lines - Google Patents

Data processing system, cache, and method of cache management including an O state for memory-consistent cache lines Download PDF

Info

Publication number
US6397303B1
US6397303B1 US09/339,408 US33940899A US6397303B1 US 6397303 B1 US6397303 B1 US 6397303B1 US 33940899 A US33940899 A US 33940899A US 6397303 B1 US6397303 B1 US 6397303B1
Authority
US
United States
Prior art keywords
cache
data
state
memory
granule
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
US09/339,408
Inventor
Ravi Kumar Arimilli
Lakshminarayana Baba Arimilli
James Stephen Fields, Jr.
Sanjeev Ghai
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
International Business Machines Corp
Original Assignee
International Business Machines Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by International Business Machines Corp filed Critical International Business Machines Corp
Priority to US09/339,408 priority Critical patent/US6397303B1/en
Assigned to INTERNATIONAL BUSINESS MACHINES CORPORATION reassignment INTERNATIONAL BUSINESS MACHINES CORPORATION ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: ARIMILLI, LAKSHMINARAYANA B., ARIMILLI, RAVI K., FIELDS, JAMES S., JR., GHAI, SANJEEV
Application granted granted Critical
Publication of US6397303B1 publication Critical patent/US6397303B1/en
Anticipated expiration legal-status Critical
Expired - Fee Related legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F12/00Accessing, addressing or allocating within memory systems or architectures
    • G06F12/02Addressing or allocation; Relocation
    • G06F12/08Addressing or allocation; Relocation in hierarchically structured memory systems, e.g. virtual memory systems
    • G06F12/0802Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches
    • G06F12/0806Multiuser, multiprocessor or multiprocessing cache systems
    • G06F12/0811Multiuser, multiprocessor or multiprocessing cache systems with multilevel cache hierarchies

Definitions

  • the present invention relates in general to data processing and, in particular, to cache management in a data processing system. Still more particularly, the present invention relates to a data processing system, cache, and method of cache management having an O state for memory-consistent. cache lines of unknown coherency.
  • a conventional multiprocessor data processing system may comprise a system bus to which a system memory and a number of processing units that each include a processor and one or more levels of cache memory are coupled. To obtain valid execution results in such a multiprocessor data processing system, a single view of the contents of memory must be provided to all of the processors by maintaining a coherent memory hierarchy.
  • a coherent memory hierarchy is maintained through the implementation of a selected coherency protocol, such as the conventional MESI protocol.
  • MESI protocol an indication of a coherency state is stored in association with each coherency granule (e.g., cache line or sector) of at least all upper level (i.e., cache) memories.
  • Each coherency granule can have one of four states, modified (M), exclusive (E), shared (S), or invalid (I), which is typically indicated by two bits in the cache directory.
  • the modified state indicates that a coherency granule is valid only in the cache storing the modified coherency granule and that the value of the modified coherency granule has not been written to (i.e., is inconsistent with) system memory.
  • coherency granule When a coherency granule is indicated as exclusive, the coherency granule is resident in, of all caches at that level of the memory hierarchy, only the cache having the coherency granule in the exclusive state. The data in the exclusive state is consistent with system memory, however. If a coherency granule is marked as shared in a cache directory, the coherency granule is resident in the associated cache and in at least one other cache at the same level of the memory hierarchy, all of the copies of the coherency granule being consistent with system memory. Finally, the invalid state generally indicates that the data and address tag associated with a coherency granule are both invalid.
  • each coherency granule is set can be dependent upon a previous state of the cache line, the type of memory access sought by processors to the associated memory address, and the state of the coherency granule in other caches. Accordingly, maintaining cache coherency in the multiprocessor data processing system requires that processors communicate messages across the system bus indicating an intention to read or write memory locations. For example, when a processing unit requires data not resident in its cache(s), the processing unit issues a read request on the system bus specifying a particular memory address. The read request is interpreted by its recipients as a request for only a single coherency granule in the lowest level cache in the processing unit.
  • the requested cache is then provided to the requester by a recipient determined by the coherency protocol, and the requestor typically caches the data in one of the valid states (i.e., M, E, or S) because of the probability that the cache line will again be accessed shortly.
  • the present invention recognizes that the conventional read request/response scenario for a multiprocessor data processing system outlined above is subject to a number of inefficiencies.
  • a significant component of the overall access latency to system memory is the internal memory latency attributable to decoding the request address and activating the appropriate word and bit lines to read out the requested cache line.
  • the requested coherency granule is only a subset of a larger data set that must be accessed at a lower level cache or system memory in order to source the requested coherency granule.
  • the internal memory latency is unnecessarily multiplied, since multiple adjacent cache lines of data could be sourced in response to a single request at approximately the same internal memory latency as a single cache line.
  • the present invention introduces the O cache consistency state, which permits unrequested memory-consistent and possibly non-coherent data to be stored in a cache, thereby reducing a processor's access latency to memory-consistent data.
  • a multiprocessor data processing system that can implement the present invention includes an interconnect, a plurality of processing units coupled to the interconnect, and at least one system memory and a plurality of caches coupled to the plurality of processing units.
  • a cache suitable for use in such a multiprocessor data processing system includes data storage containing multiple granules of data and a number of state fields associated with the granules of data. Each state field has a plurality of possible states including an O state indicating that an associated granule is consistent with corresponding data in the memory and has unknown coherency with respect to peer caches in the data processing system.
  • a cache is permitted to store memory-consistent, but possibly non-coherent data in order to offer processing units in the data processing system lower latency to an image of system memory.
  • FIG. 1 depicts an illustrative embodiment of a first multiprocessor data processing system with which the present invention may advantageously be utilized;
  • FIG. 2 is a high level block diagram of a cache in accordance with the present invention.
  • FIG. 3 is a state transition table summarizing cache state transitions, snoop responses, and combined responses for various transactions on the system interconnect of the data processing system shown in FIG. 1;
  • FIG. 4 is a block diagram depicting an illustrative embodiment of a second data processing system in accordance with the present invention, which has a hierarchical interconnect structure.
  • data processing system 8 includes a number of processing units 10 a - 10 c coupled to a system interconnect 12 .
  • Each processing unit 10 is an integrated circuit including one or more processors 14 .
  • each of processors 14 also includes an associated level one (L 1 ) cache 16 , which temporarily stores instructions and data that are likely to be accessed by the associated processor 14 .
  • L 1 caches 16 are illustrated in FIG. 1 as unified caches that store both instruction and data (both referred to hereinafter simply as data), those skilled in the art will appreciate that each of L 1 caches 16 could alternatively be implemented as bifurcated instruction and data caches.
  • the memory hierarchy of data processing system 8 also includes distributed system memories 22 , which form the lowest level of volatile data storage in the memory hierarchy, and one or more lower levels of cache memory, such as on-chip level two (L 2 ) caches 18 and off-chip L 3 caches 20 , which are utilized to stage data from system memories 22 to processors 14 .
  • L 2 on-chip level two
  • L 3 caches 20 off-chip L 3 caches 20
  • each succeeding lower level of the memory hierarchy is typically capable of storing a larger amount of data than higher levels, but at a higher access latency.
  • L 1 caches 16 may each have 512 64-byte cache lines for a total storage capacity of 32 kilobytes (kB), all at single cycle latency.
  • System interconnect 12 which can comprise one or more buses or a cross-point switch, serves as a conduit for communication transactions between processing units 10 and other snoopers (e.g., L 3 caches 20 ) coupled to system interconnect 12 .
  • a typical transaction on system interconnect 12 begins with a request, which may include a transaction field indicating the type of transaction, source and destination tags indicating the source and intended recipient(s) of the transaction, respectively, and an address and/or data.
  • Each device connected to system interconnect 12 preferably snoops all transactions on system interconnect 12 and, if appropriate, responds to the request with a snoop response.
  • response logic 24 provides a combined response indicating what action, if any, each snooper is to take in response to the request.
  • These actions may include sourcing data on system interconnect 12 , storing data provided by the requesting snooper, etc.
  • each cache line within data array 34 is sectored into two sectors 38 a , 38 b that can be individually accessed and modified.
  • the sector size utilized by each cache is the same as the cache line size of the associated higher level cache, if any.
  • L 1 caches 16 have 64-byte cache lines
  • L 2 caches 18 and L 3 caches 20 preferably implement 128-byte (two 64-byte sectors) and 256-byte (two 128-byte sectors) cache lines, respectively.
  • the cache lines stored within data array 34 are recorded in cache directory 32 , which contains one directory entry for each way in data array 34 .
  • Each directory entry comprises a tag field 40 , a status field 42 , a least recently used (LRU) field 44 , and an inclusion field 46 .
  • Tag field 40 specifies which cache line is stored in the corresponding way of data array 34 by storing the tag bits of the system memory address of the cache line.
  • status field 42 separately indicates the coherency and/or consistency status of each sector of the cache line stored in the corresponding way of data array 34 utilizing predefined bit combinations.
  • LRU field 44 indicates how recently the corresponding way of data array 34 has been accessed relative to the other ways of its congruence class, thereby indicating which cache line should be evicted from the congruence class in case of a cache miss.
  • inclusion field 46 indicates whether or not each sector of the cache line stored in the corresponding way of data array 34 is also stored in the local memory unit (i.e., cache or system memory) at the next lowest level of the memory hierarchy.
  • cache controller 36 manages storage and retrieval of data within data array 34 and updates to cache directory 32 in response to signals received from the associated components of the memory hierarchy and transactions snooped on system interconnect 12 . As illustrated, cache controller 36 maintains a read queue 50 and a write queue 52 from which cache controller 36 performs updates to cache directory 32 and accesses to data array 34 .
  • O state is defined as the state of a data granule (e.g., cache line or sector) in cache that is consistent with corresponding data in system memory but has unknown coherency with respect to at least data stored in the caches at the same level of the memory hierarchy (i.e., peer caches).
  • consistency is defined as identity of corresponding data between a cache and system memory
  • coherency is defined as knowledge of which cached copy of data associated with a particular address, if any, is the correct data.
  • a cache holding a data granule in O state has no information regarding the state of the data granule (e.g., M, E, S, or I) in remote caches.
  • the O state can be incorporated within any cache state protocol, such as the conventional MESI protocol discussed above or a variant thereof.
  • L 2 and L 3 caches 18 and 20 utilize the O state with only the M, S and I states from the MESI protocol.
  • additional states may be utilized to maintain more state information.
  • an O R state may be utilized to signify that a cache, among its peer caches in the memory hierarchy, is the cache that has most recently received O state data and is therefore responsible for sourcing the O state data granule to another cache in a cache-to-cache transfer.
  • An O M state may also or alternatively be utilized to designate a cache that has a shorter access latency to memory for a data granule than its peer caches and that is therefore responsible, among its peer caches, for sourcing the O state data granule to another cache in a cache-to-cache transfer.
  • this cache state can be explicit in a cache directory (e.g., in L 3 caches 20 ) or can be implicit in the snoop responses and combined responses of the system interconnect communication protocol.
  • an I P state may be utilized to signify that a data granule formerly held in O or O R state is known to be invalid and therefore should not be prefetched from lower level cache or memory.
  • I prefetch hint or O (or coherence of 0 read another O R if (or O R ) state sector of cache used) sector unknown line since not snooped
  • peer cache gives for sector modified snoop response O R source sector to O other cache another cache that assumes stores sector in responsibility O R state for sourcing O state sector O or O R data request by S coherence of any snooper and sector can be combined response determined from is shared or null combined snoop response O or O R any write I data in O state is an image of memory and is therefore invalid if a cache holds modified data
  • data processing system 8 implements the snoop responses shown in Table II, which lists the snoop responses in order of descending priority.
  • snoopers within data processing system 8 may also provide a prefetch flag with the snoop response indicating information regarding prefetching of the non-requested sector.
  • the prefetch flags that may be provided in or with the snoop response include NP (“no prefetch”), which indicates that the non-requested sector should not be sourced to the requestor, for example, because the snooper holds the non-requested sector in M state, and LP (“L 3 prefetch”), which the local L 3 cache provides if it holds the non-requested sector in O or S state to indicate that it will source the non-requested sector.
  • the priority of data sources for the non-requested sector are: (1) the local L 3 cache in S or O state; and (2) the L 3 cache coupled to the system memory storing the sector or the system memory itself.
  • the snoop responses are compiled by response logic 24 , which provides one of the combined responses given in Table III to all snoopers coupled to system interconnect 12 .
  • the combined response designates the snooper driving the highest priority snoop response as the source of the requested data.
  • the prefetch flags that may be passed with or in the combined response include NP, indicating that no data will be sourced for the non-requested sector, and LP, indicating that the local L 3 cache will source the non-requested sector. If neither NP or LP is asserted, data for the non-requested sector will be sourced from the L 3 cache in front of system memory, which may be required to fetch the data from system memory.
  • State transition table 60 summarizes cache state transitions, snoop responses and combined responses for various operating scenarios of data processing system 8 .
  • State transition table 60 assumes that L 2 caches 18 a - 18 c implement the M, S, I, O and O R states (the O R state is designated simply as R), that L 3 caches 20 a - 20 c implement the M, S, I, and O states, and that snoopers implement the system interconnect protocol described above.
  • Rows 62 - 80 of state transition table 60 illustrate state transitions due to requests issued by L 2 cache 18 a on system interconnect 12 to fill sector 0 with requested data identified by a memory address associated with a storage location in system memory 22 c . As shown in row 62 , if both the requested sector and the non-requested sector are invalid in all caches coupled to system interconnect 12 , all of L 2 cache 18 and L 3 caches 20 provide Null snoop responses to the request.
  • Response logic 24 compiles these Null snoop responses and issues a Null combined response, which causes system memory 22 c to source both the requested sector (sector 0 ) and the non-requested sector (sector 1 ) to L 2 cache 18 a via L 3 cache 20 c .
  • L 2 cache 18 a holds the requested sector in S state and holds the non-requested sector in O R state.
  • local L 3 cache 20 a allocates a cache line for the data and stores the requested sector and non-requested sector in S state and O state, respectively.
  • L 3 cache 20 a holds the non-requested sector in O state and L 3 cache 20 c holds the requested sector in O (or O M ) state, all snoopers drive Null snoop responses, and L 3 cache 20 a asserts the LP prefetch flag to indicate that it can source the non-requested sector.
  • Response logic 24 then compiles the Null snoop responses and issues a Null combined response with the LP prefetch flag set. This combined response instructs L 3 cache 20 c to source the requested sector (sector 0 ) and L 3 cache 20 a to source the non-requested sector (sector 1 ) to L 2 cache 18 a .
  • L 2 cache 18 a thereafter holds the requested sector in S state and holds the non-requested sector in O R state.
  • Local L 3 cache 20 a also snoops the requested sector and stores the requested sector in S state.
  • L 3 cache 20 a does not update the state of the non-requested sector, however, because the system memory address associated with the non-requested sector was not snooped.
  • Row 66 depicts a similar scenario in which L 3 cache 20 c stores the sector requested by L 2 cache 18 a in S state, and L 3 cache 20 a stores the non-requested sector in S state.
  • L 3 cache 20 c stores the sector requested by L 2 cache 18 a in S state
  • L 3 cache 20 a stores the non-requested sector in S state.
  • the same snoop responses and combined response are provided in this example, and the same caches source the requested and non-requested sectors to L 2 cache 18 a .
  • L 2 cache 18 a thereafter holds the requested sector in S state and holds the non-requested sector in O R state.
  • Local L 3 cache 20 a also snoops the requested sector and stores the requested sector in S state.
  • L 3 cache 20 a does not update the state of the non-requested sector from the S state.
  • L 3 cache 20 a holds the requested sector in O state and L 3 cache 20 c holds the non-requested sector in O state.
  • L 2 caches 18 a - 18 c and L 3 caches 20 b - 20 c all provide Null snoop responses, and L 3 cache 20 a provides an L 3 O snoop response to indicate that the local L 3 cache holds the requested sector in O state.
  • Response logic 24 compiles these snoop responses and issues an L 3 O combined response indicating that L 3 cache 20 a is to source the requested sector and L 3 cache 20 c is to source the non-requested sector.
  • L 2 cache 18 a After caching the data, L 2 cache 18 a holds the requested sector in S state and holds the non-requested sector in O R state. In addition, local L 3 cache 20 a caches the requested sector in S state and caches the non-requested sector in O state.
  • Row 70 depicts a similar scenario in which local L 3 cache 20 a stores the sector requested by L 2 cache 18 a in S state, and L 3 cache 20 c stores the non-requested sector in S state.
  • L 2 caches 18 a - 18 c and L 3 caches 20 b - 20 c all provide Null snoop responses
  • L 3 cache 20 a provides an L 3 S snoop response to indicate that the local L 3 cache holds the requested sector in S state.
  • Response logic 24 compiles these snoop responses and issues an L 3 S combined response indicating that L 3 cache 20 a is to source the requested sector and L 3 cache 20 c is to source the non-requested sector.
  • L 2 cache 18 a After caching the sourced data, L 2 cache 18 a holds the requested sector in S state and holds the non-requested sector in O R state. In addition, local L 3 cache 20 a caches the requested sector in S state and caches the non-requested sector in O state.
  • L 2 cache 18 b stores both the requested sector and the non-requested sector in O R state
  • L 3 cache 20 a stores both sectors in O state
  • L 3 cache 20 b stores the non-requested sector in M state
  • L 3 cache 20 c stores both sectors in O state.
  • L 2 caches 18 a and 18 c and L 3 cache 20 c all provide Null snoop responses
  • L 3 cache 20 b provides a Null snoop response with the NP prefetch flag set
  • L 2 cache 18 b provides an L 2 O snoop response
  • L 3 cache 20 a provides an L 3 O snoop response with the LP prefetch flag set.
  • response logic 24 provides an L 2 O combined response with the NP flag set, which instructs L 2 cache 18 b to source the requested sector and inhibits sourcing of the non-requested sector.
  • the states of the sectors in L 2 cache 18 a are S and I, respectively.
  • L 3 caches 20 a and 20 c are also able to upgrade the states of their copies of requested sector from O to S.
  • L 2 cache 18 b would source both the requested sector and the non-requested sector to L 2 cache 18 a and update the state of its copy of the non-requested sector from O R to O.
  • Row 74 of state transition table 60 illustrates an operating scenario in which L 2 cache 18 b stores both the requested sector and the non-requested sector in M state, L 3 cache 20 a stores both sectors in O state, L 3 caches 20 a and 20 c stores the requested sector in O state, and L 3 cache 20 c stores the non-requested sector in O state.
  • L 2 caches 18 a and 18 c and L 3 cache 20 b and 20 c all provide Null snoop responses
  • L 3 cache 20 b provides a MOD snoop response with the NP prefetch flag set
  • L 3 cache 20 a provides an L 3 O snoop response.
  • response logic 24 provides an MOD combined response with the NP flag set, which indicates that L 2 cache 18 b is to source the requested sector and that the non-requested sector will not be sourced.
  • the states of the sectors in L 2 cache 18 a are S and I, respectively.
  • L 3 caches 20 a and 20 c are also able to upgrade the states of their copies of requested sector from O to S.
  • Response logic 24 compiles these snoop responses and issues a Null combined response with the NP prefetch flag set, which causes L 3 cache 20 c to source the requested sector and inhibits sourcing of the non-requested sector.
  • L 2 cache 18 a stores the requested sector in S state
  • L 3 cache 20 a snoops the requested sector and also caches it in S state.
  • L 3 cache 20 c also updates the state of the requested sector from O to S.
  • Row 78 of state transition table 60 summarizes an operating scenario in which prior to the data request by L 2 cache 18 a , L 2 cache 18 b stores the requested sector in O R state, caches 18 a and 20 c store the non-requested sector in S state, and L 3 cache 18 a stores the non-requested sector in O state.
  • Any cache having the specified sector in an O state can upgrade the state of the sector to at least S state if no snooper responds with a MOD snoop response, but all snooping caches holding non-modified copies of the sector (or at least all snooping caches in an O state) must invalidate their copies if a snooper responds within a MOD snoop response.
  • row 82 summarizes an operating scenario in which L 3 cache 20 a issues an address-only query specifying the memory address in system memory 22 c associated with sector 0 , which L 3 cache 20 a holds in O state.
  • all caches issue a Null snoop response except L 2 cache 18 b , which provides an L 2 O snoop response to indicate that it holds the specified sector in O R state.
  • Response logic 24 compiles these snoop responses and issues an L 2 O combined response, which signifies to all snoopers that no snooper holds the specified sector in M state. Accordingly, both L 2 cache 18 b and L 3 cache 20 a upgrade the state of the specified sector to S state. And because the transaction code of the transaction identified it as an address-only query, neither the specified sector nor the non-specified sector is sourced on system interconnect 12 .
  • Row 84 illustrates another example of an operating scenario in which L 2 cache 18 c issues an address-only query specifying a system memory address in system memory 22 c for which L 2 cache 18 c stores associated data in sector 0 in O R state and L 2 cache 18 a and L 3 cache 20 c store associated data in S state.
  • all caches provide Null snoop responses, and response logic 24 accordingly provides a Null combined response.
  • L 2 cache 18 c upgrades the state of sector 0 from O R state to S state.
  • the requesting cache can update the state of the specified sector from an O state to E state.
  • the O R state is utilized to indicate which cache (if any), from among the caches connected to a common interconnect, is responsible for sourcing requested or prefetch data from O state.
  • the O R state is utilized to indicate which cache (if any), from among the caches connected to a common interconnect, is responsible for sourcing requested or prefetch data from O state.
  • only one cache at a particular level of the memory hierarchy can have a particular data granule stored in O R state at a time.
  • one cache on each separately snooped interconnect can have the same data stored in O R state.
  • O R designates one cache among the peer caches commonly connected to a snooped interconnect that is responsible for sourcing data in O state to other peer caches coupled to the snooped interconnect;
  • a cache holding data in O R state can also source the data to an associated higher level cache within the cache hierarchy (i.e., a higher level cache in the same cluster);
  • L 2 cache 120 a which has cache state of I/I for the sectors of a particular cache line, issues a read request for sector 0 of the cache line.
  • the cache states of L 2 caches 110 b - 110 d for the cache line in question are I/I, O R /I, and O/I, respectively, and the cache states of L 3 caches 128 a - 128 c are S/I, O/O, and O/O R , respectively. Because L 3 cache 128 a records S state for the requested sector, it is known that the requested sector in not modified in another cache.
  • L 2 cache 120 c sources the requested sector to L 2 cache 120 a and transitions to S state.
  • L 3 cache 128 a forwards the read request to lower level interconnect 130 in order to prefetch the non-requested sector (sector 1 ).
  • L 3 cache 12 c sources sector 1 and updates the state of sector 1 to O state.
  • L 3 cache 128 a updates the state of sector 1 to O R state and supplies sector 1 , the prefetch data, to L 2 cache 120 a .
  • L 2 cache 120 a then updates the states of sectors 0 and 1 to S and O R , respectively.
  • the present invention provides an improved cache state protocol that permits memory-consistent data of unknown coherency to be stored in the cache memory of a data processing system in an O state in order to reduce access latency to an image of memory.
  • the present invention permits the state of such memory-consistent data to be updated both in response to snooping request transactions issued by other caches and in response to address-only queries and conditional read transactions issued by the cache storing data in O state.
  • data held in O state may be sourced to other caches, both as requested data and as prefetch (non-requested) data through the implementation of the O R and/or O M states.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Memory System Of A Hierarchy Structure (AREA)

Abstract

A multiprocessor data processing system includes an interconnect, a plurality of processing units coupled to the interconnect, and at least one system memory and a plurality of caches coupled to the plurality of processing units. A cache suitable for use in such a data processing system includes data storage containing multiple granules of data and a number of state fields associated with the granules of data. Each state field has a plurality of possible states including an O state indicating that an associated granule is consistent with corresponding data in the memory and has unknown coherency with respect to peer caches in the data processing system. Thus, a cache is permitted to store memory-consistent, but possibly non-coherent data in order to offer processing units in the data processing system lower latency to an image of system memory.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS
This application is related to the following co-pending applications, which are filed on even date herewith and incorporated herein by reference:
(1) application Ser. No. 09/339,407;
(2) application Ser. No. 09/339,406;
(3) application Ser. No. 09/339,405; and
(4) application Ser. No. 09/339,404.
BACKGROUND OF THE INVENTION
1. Technical Field
The present invention relates in general to data processing and, in particular, to cache management in a data processing system. Still more particularly, the present invention relates to a data processing system, cache, and method of cache management having an O state for memory-consistent. cache lines of unknown coherency.
2. Description of the Related Art
A conventional multiprocessor data processing system may comprise a system bus to which a system memory and a number of processing units that each include a processor and one or more levels of cache memory are coupled. To obtain valid execution results in such a multiprocessor data processing system, a single view of the contents of memory must be provided to all of the processors by maintaining a coherent memory hierarchy.
A coherent memory hierarchy is maintained through the implementation of a selected coherency protocol, such as the conventional MESI protocol. According to the MESI protocol, an indication of a coherency state is stored in association with each coherency granule (e.g., cache line or sector) of at least all upper level (i.e., cache) memories. Each coherency granule can have one of four states, modified (M), exclusive (E), shared (S), or invalid (I), which is typically indicated by two bits in the cache directory. The modified state indicates that a coherency granule is valid only in the cache storing the modified coherency granule and that the value of the modified coherency granule has not been written to (i.e., is inconsistent with) system memory. When a coherency granule is indicated as exclusive, the coherency granule is resident in, of all caches at that level of the memory hierarchy, only the cache having the coherency granule in the exclusive state. The data in the exclusive state is consistent with system memory, however. If a coherency granule is marked as shared in a cache directory, the coherency granule is resident in the associated cache and in at least one other cache at the same level of the memory hierarchy, all of the copies of the coherency granule being consistent with system memory. Finally, the invalid state generally indicates that the data and address tag associated with a coherency granule are both invalid.
The state to which each coherency granule is set can be dependent upon a previous state of the cache line, the type of memory access sought by processors to the associated memory address, and the state of the coherency granule in other caches. Accordingly, maintaining cache coherency in the multiprocessor data processing system requires that processors communicate messages across the system bus indicating an intention to read or write memory locations. For example, when a processing unit requires data not resident in its cache(s), the processing unit issues a read request on the system bus specifying a particular memory address. The read request is interpreted by its recipients as a request for only a single coherency granule in the lowest level cache in the processing unit. The requested cache is then provided to the requester by a recipient determined by the coherency protocol, and the requestor typically caches the data in one of the valid states (i.e., M, E, or S) because of the probability that the cache line will again be accessed shortly.
The present invention recognizes that the conventional read request/response scenario for a multiprocessor data processing system outlined above is subject to a number of inefficiencies. First, given the large communication latency associated with accesses to lower levels of the memory hierarchy (particularly to system memory) in state of the art systems and the statistical likelihood that data adjacent to a requested cache line in lower level cache or system memory will subsequently be requested, it is inefficient to supply only the requested coherency granule in response to a request.
Second, a significant component of the overall access latency to system memory is the internal memory latency attributable to decoding the request address and activating the appropriate word and bit lines to read out the requested cache line. In addition, it is typically the case that the requested coherency granule is only a subset of a larger data set that must be accessed at a lower level cache or system memory in order to source the requested coherency granule. Thus, when system memory receives multiple sequential requests for adjacent cache lines, the internal memory latency is unnecessarily multiplied, since multiple adjacent cache lines of data could be sourced in response to a single request at approximately the same internal memory latency as a single cache line.
SUMMARY OF THE INVENTION
In view of the above and other shortcomings in the art recognized by the present invention, the present invention introduces the O cache consistency state, which permits unrequested memory-consistent and possibly non-coherent data to be stored in a cache, thereby reducing a processor's access latency to memory-consistent data.
A multiprocessor data processing system that can implement the present invention includes an interconnect, a plurality of processing units coupled to the interconnect, and at least one system memory and a plurality of caches coupled to the plurality of processing units. A cache suitable for use in such a multiprocessor data processing system includes data storage containing multiple granules of data and a number of state fields associated with the granules of data. Each state field has a plurality of possible states including an O state indicating that an associated granule is consistent with corresponding data in the memory and has unknown coherency with respect to peer caches in the data processing system. Thus, a cache is permitted to store memory-consistent, but possibly non-coherent data in order to offer processing units in the data processing system lower latency to an image of system memory.
All objects, features, and advantages of the present invention will become apparent in the following detailed written description.
BRIEF DESCRIPTION OF THE DRAWINGS
The novel features believed characteristic of the invention are set forth in the appended claims. The invention itself however, as well as a preferred mode of use, further objects and advantages thereof, will best be understood by reference to the following detailed description of an illustrative embodiment when read in conjunction with the accompanying drawings, wherein:
FIG. 1 depicts an illustrative embodiment of a first multiprocessor data processing system with which the present invention may advantageously be utilized;
FIG. 2 is a high level block diagram of a cache in accordance with the present invention;
FIG. 3 is a state transition table summarizing cache state transitions, snoop responses, and combined responses for various transactions on the system interconnect of the data processing system shown in FIG. 1; and
FIG. 4 is a block diagram depicting an illustrative embodiment of a second data processing system in accordance with the present invention, which has a hierarchical interconnect structure.
DETAILED DESCRIPTION OF ILLUSTRATIVE EMBODIMENT
System Architecture
With reference now to the figures and in particular with reference to FIG. 1, there is illustrated a high level block diagram of a first multiprocessor data processing system in accordance with the present invention. As depicted, data processing system 8 includes a number of processing units 10 a-10 c coupled to a system interconnect 12. Each processing unit 10 is an integrated circuit including one or more processors 14. In addition to the registers, instruction flow logic and execution units utilized to execute program instructions, each of processors 14 also includes an associated level one (L1) cache 16, which temporarily stores instructions and data that are likely to be accessed by the associated processor 14. Although L1 caches 16 are illustrated in FIG. 1 as unified caches that store both instruction and data (both referred to hereinafter simply as data), those skilled in the art will appreciate that each of L1 caches 16 could alternatively be implemented as bifurcated instruction and data caches.
As further illustrated in FIG. 1, the memory hierarchy of data processing system 8 also includes distributed system memories 22, which form the lowest level of volatile data storage in the memory hierarchy, and one or more lower levels of cache memory, such as on-chip level two (L2) caches 18 and off-chip L3 caches 20, which are utilized to stage data from system memories 22 to processors 14. As understood by those skilled in the art, each succeeding lower level of the memory hierarchy is typically capable of storing a larger amount of data than higher levels, but at a higher access latency. For example, in an exemplary embodiment, L1 caches 16 may each have 512 64-byte cache lines for a total storage capacity of 32 kilobytes (kB), all at single cycle latency. L2 caches 18 may each have 2048 128-byte cache lines for a total storage capacity of 256 kB at approximately 10 cycle latency. L3 caches 20 may each have 4096 256-byte cache lines for a total storage capacity of 1 MB, at a latency of approximately 40-60 cycles. Finally, each system memory 22 can store tens or hundreds of megabytes of data at an even longer latency, for example, 300-400 cycles. Given the large disparity in access latencies between the various levels of the memory hierarchy, it is advantageous to reduce accesses to lower levels of the memory hierarchy and, in particular, to system memories 22.
System interconnect 12, which can comprise one or more buses or a cross-point switch, serves as a conduit for communication transactions between processing units 10 and other snoopers (e.g., L3 caches 20) coupled to system interconnect 12. A typical transaction on system interconnect 12 begins with a request, which may include a transaction field indicating the type of transaction, source and destination tags indicating the source and intended recipient(s) of the transaction, respectively, and an address and/or data. Each device connected to system interconnect 12 preferably snoops all transactions on system interconnect 12 and, if appropriate, responds to the request with a snoop response. As discussed further below, such snoop responses are received and compiled by response logic 24, which provides a combined response indicating what action, if any, each snooper is to take in response to the request. These actions may include sourcing data on system interconnect 12, storing data provided by the requesting snooper, etc.
Those skilled in the art will appreciate that data processing system 8 can include many additional components, such as bridges to additional interconnects, I/O devices, non-volatile storage, ports for connection to networks or attached devices, etc. Because such additional components are not necessary for an understanding of the present invention, they are not illustrated in FIG. 1 or discussed further herein.
Cache Architecture
With reference now to FIG. 2, there is depicted a more detailed block diagram of an illustrative embodiment of a cache 30 that may be utilized to implement any of L1 caches 16, L2 caches 18 and L3 caches 20 in accordance with the present invention. In the illustrative embodiment, cache 30 is a four-way set associative cache including a directory 32, a data array 34, and a cache controller 36. Accordingly, data array 34 of cache 30 comprises a number of congruence classes that each contain four ways for storing cache lines. As in conventional set associative caches, memory locations in system memories 22 are mapped to particular congruence classes within data array 34 utilizing predetermined index bits within the system memory address.
As further shown in FIG. 2, each cache line within data array 34 is sectored into two sectors 38 a, 38 b that can be individually accessed and modified. Although not required by the present invention, it is convenient if the sector size utilized by each cache is the same as the cache line size of the associated higher level cache, if any. For example, if L1 caches 16 have 64-byte cache lines, L2 caches 18 and L3 caches 20 preferably implement 128-byte (two 64-byte sectors) and 256-byte (two 128-byte sectors) cache lines, respectively.
The cache lines stored within data array 34 are recorded in cache directory 32, which contains one directory entry for each way in data array 34. Each directory entry comprises a tag field 40, a status field 42, a least recently used (LRU) field 44, and an inclusion field 46. Tag field 40 specifies which cache line is stored in the corresponding way of data array 34 by storing the tag bits of the system memory address of the cache line. As discussed in detail below, status field 42 separately indicates the coherency and/or consistency status of each sector of the cache line stored in the corresponding way of data array 34 utilizing predefined bit combinations. LRU field 44 indicates how recently the corresponding way of data array 34 has been accessed relative to the other ways of its congruence class, thereby indicating which cache line should be evicted from the congruence class in case of a cache miss. Finally, inclusion field 46 indicates whether or not each sector of the cache line stored in the corresponding way of data array 34 is also stored in the local memory unit (i.e., cache or system memory) at the next lowest level of the memory hierarchy.
Still referring to FIG. 2, cache controller 36 manages storage and retrieval of data within data array 34 and updates to cache directory 32 in response to signals received from the associated components of the memory hierarchy and transactions snooped on system interconnect 12. As illustrated, cache controller 36 maintains a read queue 50 and a write queue 52 from which cache controller 36 performs updates to cache directory 32 and accesses to data array 34.
Cache State Protocol
In order to reduce high latency accesses to system memories 22, the present invention introduces the O state into the cache state protocol implemented by cache controllers 36, where O state is defined as the state of a data granule (e.g., cache line or sector) in cache that is consistent with corresponding data in system memory but has unknown coherency with respect to at least data stored in the caches at the same level of the memory hierarchy (i.e., peer caches). In this definition, consistency is defined as identity of corresponding data between a cache and system memory, and coherency is defined as knowledge of which cached copy of data associated with a particular address, if any, is the correct data. Thus, a cache holding a data granule in O state has no information regarding the state of the data granule (e.g., M, E, S, or I) in remote caches.
The O state can be incorporated within any cache state protocol, such as the conventional MESI protocol discussed above or a variant thereof. In a minimal implementation of the O state in data processing system 8, L2 and L3 caches 18 and 20 utilize the O state with only the M, S and I states from the MESI protocol. In other embodiments, additional states may be utilized to maintain more state information. For example, an OR state may be utilized to signify that a cache, among its peer caches in the memory hierarchy, is the cache that has most recently received O state data and is therefore responsible for sourcing the O state data granule to another cache in a cache-to-cache transfer. An OM state may also or alternatively be utilized to designate a cache that has a shorter access latency to memory for a data granule than its peer caches and that is therefore responsible, among its peer caches, for sourcing the O state data granule to another cache in a cache-to-cache transfer. As discussed below, this cache state can be explicit in a cache directory (e.g., in L3 caches 20) or can be implicit in the snoop responses and combined responses of the system interconnect communication protocol. In addition, an IP state may be utilized to signify that a data granule formerly held in O or OR state is known to be invalid and therefore should not be prefetched from lower level cache or memory. A summary of state transitions for the O, OR, and IP cache states is given below in Table I. For operations and state transitions not listed in Table I, the IP state otherwise behaves and interacts with other states like the conventional MESI Invalid (I) state.
TABLE I
Initial Next
state Operation state Comments
I prefetch hint or O (or coherence of 0
read another OR if (or OR) state
sector of cache used) sector unknown
line since not snooped
IP prefetch hint or IP no prefetch
read another performed due to
sector of cache cached modified
line data
O or OR read request by IP IP inhibits
any snooper and future prefetches
peer cache gives for sector
modified snoop
response
OR source sector to O other cache
another cache that assumes
stores sector in responsibility
OR state for sourcing O
state sector
O or OR data request by S coherence of
any snooper and sector can be
combined response determined from
is shared or null combined snoop
response
O or OR any write I data in O state
is an image of
memory and is
therefore invalid
if a cache holds
modified data
System Interconnect Communication Protocol
The snoop responses and combined responses utilized to govern communication of data on system interconnect 12 depend upon the cache state protocol implemented by L2 cache 18 and L3 cache 20. Assuming that L2 caches 18 and L3 caches 20 have sectored cache lines containing two sectors each (as shown in FIG. 2), that separate cache states are maintained for each sector, and that of the various possible cache line states, L2 caches 18 a-18 c implement the M, S, I, O and OR states, and L3 caches 20 a-20 c implement the M, S, I, and O states, the relative priorities of data sources for a request issued on interconnect 12 are as follows (from highest to lowest):
(1) an L2 cache or L3 cache holding the requested sector in M state;
(2) the L3 cache beneath the requester (i.e., the local L3 cache) holding the requested sector in S state;
(3) an L2 cache holding the requested sector in OR state;
(4) the local L3 cache holding the requested sector in O state;
(5) the L3 cache coupled to the system memory storing the requested sector (in S or O state); and
(6) system memory.
In order to implement this priority scheme, data processing system 8 implements the snoop responses shown in Table II, which lists the snoop responses in order of descending priority.
TABLE II
Snoop
response Comment
Retry retry the request
MOD driven by an L2 or L3 cache holding
requested sector in M state
L3S driven by the local L3 cache if it holds
the requested sector in S state
L2O driven by an L2 cache holding the requested
sector in OR state
L3O driven by the local L3 cache if it holds
the requested sector in O state
Null default response of any snooper
In addition to providing one of the snoop responses given in Table II in response to each request, snoopers within data processing system 8 may also provide a prefetch flag with the snoop response indicating information regarding prefetching of the non-requested sector. In the embodiment of FIG. 3, the prefetch flags that may be provided in or with the snoop response include NP (“no prefetch”), which indicates that the non-requested sector should not be sourced to the requestor, for example, because the snooper holds the non-requested sector in M state, and LP (“L3 prefetch”), which the local L3 cache provides if it holds the non-requested sector in O or S state to indicate that it will source the non-requested sector. Thus, the priority of data sources for the non-requested sector are: (1) the local L3 cache in S or O state; and (2) the L3 cache coupled to the system memory storing the sector or the system memory itself.
As noted above, the snoop responses (including prefetch flag(s), if any) are compiled by response logic 24, which provides one of the combined responses given in Table III to all snoopers coupled to system interconnect 12. In general, the combined response designates the snooper driving the highest priority snoop response as the source of the requested data.
TABLE III
Combined
response Comment
Retry requestor must retry the request
MOD L2 or L3 cache holding requested sector in
M state will source data
L3S local L3 cache holding the requested sector
in S state will source data
L2O L2 cache holding the requested sector in OR
state will source data
L3O local L3 cache holding the requested sector
in O state will source data
Null data will be sourced by the L3 cache in
front of system memory from S or O state or
from the system memory itself
The prefetch flags that may be passed with or in the combined response include NP, indicating that no data will be sourced for the non-requested sector, and LP, indicating that the local L3 cache will source the non-requested sector. If neither NP or LP is asserted, data for the non-requested sector will be sourced from the L3 cache in front of system memory, which may be required to fetch the data from system memory.
Referring now to FIG. 3, a state transition table 60 is given that summarizes cache state transitions, snoop responses and combined responses for various operating scenarios of data processing system 8. State transition table 60 assumes that L2 caches 18 a-18 c implement the M, S, I, O and OR states (the OR state is designated simply as R), that L3 caches 20 a-20 c implement the M, S, I, and O states, and that snoopers implement the system interconnect protocol described above.
Rows 62-80 of state transition table 60 illustrate state transitions due to requests issued by L2 cache 18 a on system interconnect 12 to fill sector 0 with requested data identified by a memory address associated with a storage location in system memory 22 c. As shown in row 62, if both the requested sector and the non-requested sector are invalid in all caches coupled to system interconnect 12, all of L2 cache 18 and L3 caches 20 provide Null snoop responses to the request. Response logic 24 compiles these Null snoop responses and issues a Null combined response, which causes system memory 22 c to source both the requested sector (sector 0) and the non-requested sector (sector 1) to L2 cache 18 a via L3 cache 20 c. L2 cache 18 a holds the requested sector in S state and holds the non-requested sector in OR state. In addition, local L3 cache 20 a allocates a cache line for the data and stores the requested sector and non-requested sector in S state and O state, respectively.
If, as shown in row 64, L3 cache 20 a holds the non-requested sector in O state and L3 cache 20 c holds the requested sector in O (or OM) state, all snoopers drive Null snoop responses, and L3 cache 20 a asserts the LP prefetch flag to indicate that it can source the non-requested sector. Response logic 24 then compiles the Null snoop responses and issues a Null combined response with the LP prefetch flag set. This combined response instructs L3 cache 20 c to source the requested sector (sector 0) and L3 cache 20 a to source the non-requested sector (sector 1) to L2 cache 18 a. L2 cache 18 a thereafter holds the requested sector in S state and holds the non-requested sector in OR state. Local L3 cache 20 a also snoops the requested sector and stores the requested sector in S state. L3 cache 20 a does not update the state of the non-requested sector, however, because the system memory address associated with the non-requested sector was not snooped.
Row 66 depicts a similar scenario in which L3 cache 20 c stores the sector requested by L2 cache 18 a in S state, and L3 cache 20 a stores the non-requested sector in S state. The same snoop responses and combined response are provided in this example, and the same caches source the requested and non-requested sectors to L2 cache 18 a. L2 cache 18 a thereafter holds the requested sector in S state and holds the non-requested sector in OR state. Local L3 cache 20 a also snoops the requested sector and stores the requested sector in S state. L3 cache 20 a does not update the state of the non-requested sector from the S state.
Referring now to row 68, an operating scenario is given in which L3 cache 20 a holds the requested sector in O state and L3 cache 20 c holds the non-requested sector in O state. In response to the request by L2 cache 18 a, L2 caches 18 a-18 c and L3 caches 20 b-20 c all provide Null snoop responses, and L3 cache 20 a provides an L3O snoop response to indicate that the local L3 cache holds the requested sector in O state. Response logic 24 compiles these snoop responses and issues an L3O combined response indicating that L3 cache 20 a is to source the requested sector and L3 cache 20 c is to source the non-requested sector. After caching the data, L2 cache 18 a holds the requested sector in S state and holds the non-requested sector in OR state. In addition, local L3 cache 20 a caches the requested sector in S state and caches the non-requested sector in O state.
Row 70 depicts a similar scenario in which local L3 cache 20 a stores the sector requested by L2 cache 18 a in S state, and L3 cache 20 c stores the non-requested sector in S state. In response to the request by L2 cache 18 a, L2 caches 18 a-18 c and L3 caches 20 b-20 c all provide Null snoop responses, and L3 cache 20 a provides an L3S snoop response to indicate that the local L3 cache holds the requested sector in S state. Response logic 24 compiles these snoop responses and issues an L3S combined response indicating that L3 cache 20 a is to source the requested sector and L3 cache 20 c is to source the non-requested sector. After caching the sourced data, L2 cache 18 a holds the requested sector in S state and holds the non-requested sector in OR state. In addition, local L3 cache 20 a caches the requested sector in S state and caches the non-requested sector in O state.
With reference now to row 72, an operating scenario is given in which L2 cache 18 b stores both the requested sector and the non-requested sector in OR state, L3 cache 20 a stores both sectors in O state, L3 cache 20 b stores the non-requested sector in M state, and L3 cache 20 c stores both sectors in O state. As shown, L2 caches 18 a and 18 c and L3 cache 20 c all provide Null snoop responses, L3 cache 20 b provides a Null snoop response with the NP prefetch flag set, L2 cache 18 b provides an L2O snoop response, and L3 cache 20 a provides an L3O snoop response with the LP prefetch flag set. Because L2 cache 18 b is the most preferred source of the requested sector and a modified copy of the non-requested sector exists, as evidenced by the assertion of the NP prefetch flag in a snoop response, response logic 24 provides an L2O combined response with the NP flag set, which instructs L2 cache 18 b to source the requested sector and inhibits sourcing of the non-requested sector. After the requested sector is sourced to L2 cache 18 a, the states of the sectors in L2 cache 18 a are S and I, respectively. L3 caches 20 a and 20 c are also able to upgrade the states of their copies of requested sector from O to S.
If the scenario shown in row 72 were altered such that L3 cache 20 b did not hold a valid copy of the non-requested sector and the communication protocol permitted L2 caches 18 to source non-requested sectors (e.g., through the inclusion of an L2P prefetch flag having higher priority than the LP prefetch flag), L2 cache 18 b would source both the requested sector and the non-requested sector to L2 cache 18 a and update the state of its copy of the non-requested sector from OR to O.
Row 74 of state transition table 60 illustrates an operating scenario in which L2 cache 18 b stores both the requested sector and the non-requested sector in M state, L3 cache 20 a stores both sectors in O state, L3 caches 20 a and 20 c stores the requested sector in O state, and L3 cache 20 c stores the non-requested sector in O state. As depicted, L2 caches 18 a and 18 c and L3 cache 20 b and 20 c all provide Null snoop responses, L3 cache 20 b provides a MOD snoop response with the NP prefetch flag set, and L3 cache 20 a provides an L3O snoop response. In response to these snoop responses, response logic 24 provides an MOD combined response with the NP flag set, which indicates that L2 cache 18 b is to source the requested sector and that the non-requested sector will not be sourced. After the requested sector is sourced to L2 cache 18 a, the states of the sectors in L2 cache 18 a are S and I, respectively. L3 caches 20 a and 20 c are also able to upgrade the states of their copies of requested sector from O to S.
Referring now to row 76 of state transition table 60, an operating scenario is given in which requesting L2 cache 18 a stores the non-requested sector in O state, and L3 cache 20 c stores both the requested sector and the non-requested sector in O state. As shown, in response to the request by L2 cache 18 a, L2 caches 18 b-18 c and L3 caches 20 a-20 c provide Null snoop responses, and L2 cache 18 a provides a Null snoop response with the NP prefetch flag set to inhibit prefetching of the non-requested sector because L2 cache 18 a already stores an image of system memory in that sector. Response logic 24 compiles these snoop responses and issues a Null combined response with the NP prefetch flag set, which causes L3 cache 20 c to source the requested sector and inhibits sourcing of the non-requested sector. At the conclusion of the transaction, L2 cache 18 a stores the requested sector in S state, and L3 cache 20 a snoops the requested sector and also caches it in S state. L3 cache 20 c also updates the state of the requested sector from O to S.
Row 78 of state transition table 60 summarizes an operating scenario in which prior to the data request by L2 cache 18 a, L2 cache 18 b stores the requested sector in OR state, caches 18 a and 20 c store the non-requested sector in S state, and L3 cache 18 a stores the non-requested sector in O state. In response to the request by L2 cache 18 a, L2 cache 18 c and L3 caches 20 b and 20 c issue Null snoop responses, L2 cache 18 a issues a Null response with the NP flag set since it already stores the non-requested sector in S state, L2 cache 18 b issues an L2O snoop response, and L3 cache 20 a issues a Null snoop response with the LP prefetch flag set. Response logic 24 compiles these snoop responses and issues an L2O combined response with the NP prefetch flag set. As a result, L2 cache 18 b sources the requested sector to L2 cache 18 a, and prefetching of the non-requested sector is inhibited. Thereafter, L2 caches stores the requested sector in S state, local L3 cache 20 a allocates the requested sector and stores it in S state, and L2 cache 18 b updates the state of its copy of the requested sector from OR to S. A similar operating scenario is given in row 80, which differs only in that L2 cache 18 a inhibits prefetching of the non-requested sector since it stores it in M state, and local L3 cache 20 a sources the requested sector, which L3 cache 20 a stores in S state, to L2 cache 18 a. Accordingly, in this example, only the state of the requested sector transitions (from I to S state).
Referring now to rows 82-86 of state transition table 60, three operating scenarios are given in which a cache holding a sector in an O state (i.e., O state or a variant thereof) updates the state of the sector through issuing an address-only query on system interconnect 12. Such address-only queries would preferably be very low-priority transactions that are utilized to update cache states during periods of low cache activity (and low bus demand if system interconnect 12 is a shared bus). Any cache having the specified sector in an O state can upgrade the state of the sector to at least S state if no snooper responds with a MOD snoop response, but all snooping caches holding non-modified copies of the sector (or at least all snooping caches in an O state) must invalidate their copies if a snooper responds within a MOD snoop response.
As illustrated, row 82 summarizes an operating scenario in which L3 cache 20 a issues an address-only query specifying the memory address in system memory 22 c associated with sector 0, which L3 cache 20 a holds in O state. In response to the address-only query, all caches issue a Null snoop response except L2 cache 18 b, which provides an L2O snoop response to indicate that it holds the specified sector in OR state. Response logic 24 compiles these snoop responses and issues an L2O combined response, which signifies to all snoopers that no snooper holds the specified sector in M state. Accordingly, both L2 cache 18 b and L3 cache 20 a upgrade the state of the specified sector to S state. And because the transaction code of the transaction identified it as an address-only query, neither the specified sector nor the non-specified sector is sourced on system interconnect 12.
Row 84 illustrates another example of an operating scenario in which L2 cache 18 c issues an address-only query specifying a system memory address in system memory 22 c for which L2 cache 18 c stores associated data in sector 0 in OR state and L2 cache 18 a and L3 cache 20 c store associated data in S state. In response to this address-only query, all caches provide Null snoop responses, and response logic 24 accordingly provides a Null combined response. As a result, L2 cache 18 c upgrades the state of sector 0 from OR state to S state.
Referring now to row 86, an exemplary operating scenario is given in which L3 cache 20 a issues an address-only query specifying a memory address in system memory 22 c for which L3 cache 20 a stores data in O state, L2 cache 18 b stores data in M state, and L3 cache 20 c stores data in S state. As shown, L2 cache 22 b issues a MOD snoop response, and all other snoopers issue Null snoop responses. Accordingly, response logic 24 issues a MOD combined response, and L3 caches 20 a and 20 c both downgrade the states of their copies of the specified sector to I state.
Although not illustrated in state transition table 60, in embodiments in which the system interconnect communication protocol provides a Null combined response to an address-only query only when peer caches of the requesting cache do not have a copy of the specified sector, the requesting cache can update the state of the specified sector from an O state to E state.
Hierarchical Implementation of OR State
As noted above, the OR state is utilized to indicate which cache (if any), from among the caches connected to a common interconnect, is responsible for sourcing requested or prefetch data from O state. As shown in the exemplary processing scenarios outlined above, only one cache at a particular level of the memory hierarchy can have a particular data granule stored in OR state at a time. However, when a data processing system includes multiple snooped interconnects, one cache on each separately snooped interconnect can have the same data stored in OR state.
For example, with reference now to FIG. 4, there is illustrated a block diagram of a second data processing system in accordance with the present invention, which has a hierarchical interconnect structure and can support multiple caches concurrently storing the same data granule in OR state. As depicted, data processing system 100 includes three clusters 102 a-102 c, which are each connected to a lower level interconnect 130 to which system memory 132 and other interconnects and devices may be coupled. Clusters 102 a-102 c are of identical construction, and each contains an upper level interconnect 126 to which four processing units (respectively identified by reference numerals 110 a-110 d, 112 a-112 d, and 114 a-114 d) and an L3 cache 128 are connected. As with processing units 10 of FIG. 1, the processing units depicted in FIG. 4 each contain an L2 cache.
In a preferred embodiment, the OR state is implemented hierarchically with the following rules:
(1) OR designates one cache among the peer caches commonly connected to a snooped interconnect that is responsible for sourcing data in O state to other peer caches coupled to the snooped interconnect;
(2) a cache holding data in OR state can also source the data to an associated higher level cache within the cache hierarchy (i.e., a higher level cache in the same cluster); and
(3) the most recent recipient of prefetch data enters OR state and must thereafter transition to O state when another peer cache enters OR state.
As an example of the operation of OR state in a hierarchical implementation, assume that L2 cache 120 a, which has cache state of I/I for the sectors of a particular cache line, issues a read request for sector 0 of the cache line. The cache states of L2 caches 110 b-110 d for the cache line in question are I/I, OR/I, and O/I, respectively, and the cache states of L3 caches 128 a-128 c are S/I, O/O, and O/OR, respectively. Because L3 cache 128 a records S state for the requested sector, it is known that the requested sector in not modified in another cache. Therefore, L2 cache 120 c sources the requested sector to L2 cache 120 a and transitions to S state. In addition, L3 cache 128 a forwards the read request to lower level interconnect 130 in order to prefetch the non-requested sector (sector 1). In response to snooping the read request, L3 cache 12 c sources sector 1 and updates the state of sector 1 to O state. L3 cache 128 a updates the state of sector 1 to OR state and supplies sector 1, the prefetch data, to L2 cache 120 a. L2 cache 120 a then updates the states of sectors 0 and 1 to S and OR, respectively.
As has been described, the present invention provides an improved cache state protocol that permits memory-consistent data of unknown coherency to be stored in the cache memory of a data processing system in an O state in order to reduce access latency to an image of memory. Advantageously, the present invention permits the state of such memory-consistent data to be updated both in response to snooping request transactions issued by other caches and in response to address-only queries and conditional read transactions issued by the cache storing data in O state. In addition, data held in O state may be sourced to other caches, both as requested data and as prefetch (non-requested) data through the implementation of the OR and/or OM states.
While the invention has been particularly shown and described with reference to a preferred embodiment, it will be understood by those skilled in the art that various changes in form and detail may be made therein without departing from the spirit and scope of the invention.

Claims (17)

What is claimed is:
1. A cache for use in a data processing system having a memory and a peer cache, said cache comprising:
data storage including a plurality of granules of data; and
a plurality of state fields associated with said plurality of granules of data, wherein each state field among said plurality of state fields has a plurality of possible states including a first state indicating that an associated granule is consistent with corresponding data in the memory and has unknown coherency with respect to a corresponding valid data granule held in said peer cache.
2. A processing unit for use in a data processing system having a memory and a peer cache, comprising:
a processor core; and
a cache hierarchy including at least one cache, said cache including:
data storage including a plurality of granules of data; and
a plurality of state fields associated with said plurality of granules of data, wherein each state field among said plurality of state fields has a plurality of possible states including a first state indicating that an associated granule is consistent with corresponding data in the memory and has unknown coherency with respect to a corresponding valid data granule held in the peer cache.
3. A data processing system, comprising:
an interconnect and a plurality of processing units coupled to said interconnect;
at least one system memory coupled to said plurality of processing units;
a plurality of caches coupled to said plurality of processing units, wherein a cache among said plurality of caches includes:
data storage including a plurality of granules of data; and
a plurality of state fields associated with said plurality of granules of data, wherein each state field among said plurality of state fields has a plurality of possible states including a first state indicating that an associated granule is consistent with corresponding data in said system memory and has unknown coherency with respect to a corresponding valid data granule held in another of said plurality of caches.
4. A method of tracking status of data in a cache of a data processing system having a memory and a plurality of caches, said method comprising:
storing a plurality of granules of data in a cache among a plurality of caches;
storing a corresponding valid data granule in a peer cache among said plurality of caches; and
setting a state field associated with a first granule among said plurality of granules to a first state among a plurality of possible states to indicate that said first granule is consistent with corresponding data in the memory and has unknown coherency with respect to said corresponding valid data granule held in said peer cache.
5. The cache of claim 1, said plurality of states including at least a second state indicating that an assocaited granule of data is inconsistent with corresponding data in the memory and a third state indicating that an associated granule of data is coherent with corresponding data in the peer cache.
6. The cache of claim 5, wherein said third state further indicates that said associated granule of data is consistent with corresponding data in the memory.
7. The cache of claim 1, wherein each of said plurality of granules of data is a sector of a cache line.
8. The cache of claim 1, wherein said cache, the memory and the peer cache belong to a memory hierarchy having at least one lower level, said cache further comprising a cache controller that sets each of said plurality of state fields to one of said plurality of states, wherein said cache controller sets a state field among said plurality of state fields to said first state in response to said cache loading, into an associated data granule, data supplied from said lower level that is not required by said cache to satisfy a data request.
9. The data processing system of claim 2, said plurality of states including at least a second state indicating that an associated granule of data is inconsistent with corresponding data in said system memory and a third state indicating that an associated granule of data is coherent with corresponding data in others of said plurality of caches.
10. The data processing system of claim 9, wherein said third state further indicates that said associated granule of data is consistent with corresponding data in said system memory.
11. The data processing system of claim 2, wherein each of said plurality of granules of data is a sector of a cache line.
12. The data processing system of claim 2, wherein said plurality of caches and said at least one system memory belong to a memory hierarchy having at least one lower level below said cache, said cache futher comprising a cache controller that sets each of said plurality of state fields to one of said plurality of states, wherein said cache controller sets a state field among said plurality of state fields to said first state in response to said cache loading, into an associated data granule, data supplied from said lower level that is not required by said cache to satisfy a data request.
13. The data processing system of claim 12, said memory hierarchy including at least one upper level above said cache.
14. The method of claim 4, and further comprising:
setting a state field associated with a second granule among said plurality of granules to a second state to indicate that said second granule of data is inconsistent with corresponding data in the memory; and
setting a state field associated with a third granule among said plurality of granules to a third state to indicate that said third granule is coherent with corresponding data in peer caches among the plurality of caches.
15. The method of claim 14, wherein setting said state field associated with said third granule to said third state further indicates that said associated granule of data is consistent with corresponding data in the memory.
16. The method of claim 4, wherein storing a plurality of granules of data in said cache comprises storing each of said plurality of granules of data in a sector of a cache line.
17. The method of claim 4, wherein said plurality of caches and memory belong to a memory hierarchy having at least one lower level, wherein said setting step comprises setting a state field among said plurality of state fields to said first state in response to said cache loading, into said first granule, data supplied from said lower level that is not required by said cache to satisfy a data request.
US09/339,408 1999-06-24 1999-06-24 Data processing system, cache, and method of cache management including an O state for memory-consistent cache lines Expired - Fee Related US6397303B1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US09/339,408 US6397303B1 (en) 1999-06-24 1999-06-24 Data processing system, cache, and method of cache management including an O state for memory-consistent cache lines

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
US09/339,408 US6397303B1 (en) 1999-06-24 1999-06-24 Data processing system, cache, and method of cache management including an O state for memory-consistent cache lines

Publications (1)

Publication Number Publication Date
US6397303B1 true US6397303B1 (en) 2002-05-28

Family

ID=23328873

Family Applications (1)

Application Number Title Priority Date Filing Date
US09/339,408 Expired - Fee Related US6397303B1 (en) 1999-06-24 1999-06-24 Data processing system, cache, and method of cache management including an O state for memory-consistent cache lines

Country Status (1)

Country Link
US (1) US6397303B1 (en)

Cited By (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20030115423A1 (en) * 1999-12-22 2003-06-19 Intel Corporation Cache states for multiprocessor cache coherency protocols
US20030204742A1 (en) * 2002-04-29 2003-10-30 Microsoft Corporation Peer-to-peer name resolution protocol (PNRP) security infrastructure and method
US20040103251A1 (en) * 2002-11-26 2004-05-27 Mitchell Alsup Microprocessor including a first level cache and a second level cache having different cache line sizes
US20050177715A1 (en) * 2004-02-09 2005-08-11 Microsoft Corporation Method and system for managing identities in a peer-to-peer networking environment
US20050183138A1 (en) * 2004-02-13 2005-08-18 Microsoft Corporation System and method for protecting a computing device from computer exploits delivered over a networked environment in a secured communication
US20050182949A1 (en) * 2004-02-13 2005-08-18 Microsoft Corporation System and method for securing a computer system connected to a network from attacks
US20060005013A1 (en) * 2004-06-30 2006-01-05 Microsoft Corporation Call signs
US20060020807A1 (en) * 2003-03-27 2006-01-26 Microsoft Corporation Non-cryptographic addressing
US20060020758A1 (en) * 2004-07-21 2006-01-26 Wheeler Andrew R System and method to facilitate reset in a computer system
US20060064754A1 (en) * 2004-02-13 2006-03-23 Microsoft Corporation Distributed network security service
US20060095965A1 (en) * 2004-10-29 2006-05-04 Microsoft Corporation Network security device and method for protecting a computing device in a networked environment
US20060224833A1 (en) * 2005-03-31 2006-10-05 International Business Machines Corporation Data processing system, cache system and method for issuing a request on an interconnect fabric without reference to a lower level cache based upon a tagged cache state
US20070250700A1 (en) * 2006-04-21 2007-10-25 Microsoft Corporation Peer-to-peer contact exchange

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4847804A (en) * 1985-02-05 1989-07-11 Digital Equipment Corporation Apparatus and method for data copy consistency in a multi-cache data processing unit
US5671391A (en) * 1994-01-10 1997-09-23 Ncr Corporation Coherent copyback protocol for multi-level cache memory systems

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4847804A (en) * 1985-02-05 1989-07-11 Digital Equipment Corporation Apparatus and method for data copy consistency in a multi-cache data processing unit
US5671391A (en) * 1994-01-10 1997-09-23 Ncr Corporation Coherent copyback protocol for multi-level cache memory systems

Cited By (38)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6694409B2 (en) * 1999-12-22 2004-02-17 Intel Corporation Cache states for multiprocessor cache coherency protocols
US20030115423A1 (en) * 1999-12-22 2003-06-19 Intel Corporation Cache states for multiprocessor cache coherency protocols
US20080295170A1 (en) * 2002-04-29 2008-11-27 Microsoft Corporation Peer-to-peer name resolution protocol (pnrp) security infrastructure and method
US7418479B2 (en) 2002-04-29 2008-08-26 Microsoft Corporation Peer-to-peer name resolution protocol (PNRP) security infrastructure and method
US7251694B2 (en) 2002-04-29 2007-07-31 Microsoft Corporation Peer-to peer name resolution protocol (PNRP) security infrastructure and method
US7680930B2 (en) 2002-04-29 2010-03-16 Microsoft Corporation Peer-to-peer name resolution protocol (PNRP) security infrastructure and method
US20060179139A1 (en) * 2002-04-29 2006-08-10 Microsoft Corporation Peer-to-peer name resolution protocol (PNRP) security infrastructure and method
US20090006849A1 (en) * 2002-04-29 2009-01-01 Microsoft Corporation Peer-to-peer name resolution protocol (pnrp) security infrastructure and method
US20030204742A1 (en) * 2002-04-29 2003-10-30 Microsoft Corporation Peer-to-peer name resolution protocol (PNRP) security infrastructure and method
US7444372B2 (en) 2002-04-29 2008-10-28 Microsoft Corporation Peer-to-peer name resolution protocol (PNRP) security infrastructure and method
US7720962B2 (en) 2002-04-29 2010-05-18 Microsoft Corporation Peer-to-peer name resolution protocol (PNRP) security infrastructure and method
US20070168512A1 (en) * 2002-04-29 2007-07-19 Microsoft Corporation Peer-to-peer name resolution protocol (PNRP) security infrastructure and method
US20060174005A1 (en) * 2002-04-29 2006-08-03 Microsoft Corporation Peer-to-peer name resolution protocol (PNRP) security infrastructure and method
US7725567B2 (en) 2002-04-29 2010-05-25 Microsoft Corporation Peer-to-peer name resolution protocol (PNRP) security infrastructure and method
US7051102B2 (en) * 2002-04-29 2006-05-23 Microsoft Corporation Peer-to-peer name resolution protocol (PNRP) security infrastructure and method
US20060161657A1 (en) * 2002-04-29 2006-07-20 Microsoft Corporation Peer-to peer name resolution protocol (PNRP) security infrastructure and method
WO2004049170A3 (en) * 2002-11-26 2006-05-11 Advanced Micro Devices Inc Microprocessor including a first level cache and a second level cache having different cache line sizes
US20040103251A1 (en) * 2002-11-26 2004-05-27 Mitchell Alsup Microprocessor including a first level cache and a second level cache having different cache line sizes
WO2004049170A2 (en) * 2002-11-26 2004-06-10 Advanced Micro Devices, Inc. Microprocessor including a first level cache and a second level cache having different cache line sizes
US20060020807A1 (en) * 2003-03-27 2006-01-26 Microsoft Corporation Non-cryptographic addressing
US8261062B2 (en) 2003-03-27 2012-09-04 Microsoft Corporation Non-cryptographic addressing
US20050177715A1 (en) * 2004-02-09 2005-08-11 Microsoft Corporation Method and system for managing identities in a peer-to-peer networking environment
US20060064754A1 (en) * 2004-02-13 2006-03-23 Microsoft Corporation Distributed network security service
US7716726B2 (en) 2004-02-13 2010-05-11 Microsoft Corporation System and method for protecting a computing device from computer exploits delivered over a networked environment in a secured communication
US7814543B2 (en) 2004-02-13 2010-10-12 Microsoft Corporation System and method for securing a computer system connected to a network from attacks
US20050182949A1 (en) * 2004-02-13 2005-08-18 Microsoft Corporation System and method for securing a computer system connected to a network from attacks
US20050183138A1 (en) * 2004-02-13 2005-08-18 Microsoft Corporation System and method for protecting a computing device from computer exploits delivered over a networked environment in a secured communication
US7603716B2 (en) 2004-02-13 2009-10-13 Microsoft Corporation Distributed network security service
US20060005013A1 (en) * 2004-06-30 2006-01-05 Microsoft Corporation Call signs
US7929689B2 (en) 2004-06-30 2011-04-19 Microsoft Corporation Call signs
US20060020758A1 (en) * 2004-07-21 2006-01-26 Wheeler Andrew R System and method to facilitate reset in a computer system
US7500056B2 (en) * 2004-07-21 2009-03-03 Hewlett-Packard Development Company, L.P. System and method to facilitate reset in a computer system
US7716727B2 (en) 2004-10-29 2010-05-11 Microsoft Corporation Network security device and method for protecting a computing device in a networked environment
US20060095965A1 (en) * 2004-10-29 2006-05-04 Microsoft Corporation Network security device and method for protecting a computing device in a networked environment
US20060224833A1 (en) * 2005-03-31 2006-10-05 International Business Machines Corporation Data processing system, cache system and method for issuing a request on an interconnect fabric without reference to a lower level cache based upon a tagged cache state
US7536513B2 (en) * 2005-03-31 2009-05-19 International Business Machines Corporation Data processing system, cache system and method for issuing a request on an interconnect fabric without reference to a lower level cache based upon a tagged cache state
US8086842B2 (en) 2006-04-21 2011-12-27 Microsoft Corporation Peer-to-peer contact exchange
US20070250700A1 (en) * 2006-04-21 2007-10-25 Microsoft Corporation Peer-to-peer contact exchange

Similar Documents

Publication Publication Date Title
US6405290B1 (en) Multiprocessor system bus protocol for O state memory-consistent data
US6345341B1 (en) Method of cache management for dynamically disabling O state memory-consistent data
US6289420B1 (en) System and method for increasing the snoop bandwidth to cache tags in a multiport cache memory subsystem
US6192451B1 (en) Cache coherency protocol for a data processing system including a multi-level memory hierarchy
US7536513B2 (en) Data processing system, cache system and method for issuing a request on an interconnect fabric without reference to a lower level cache based upon a tagged cache state
US7032074B2 (en) Method and mechanism to use a cache to translate from a virtual bus to a physical bus
US8495308B2 (en) Processor, data processing system and method supporting a shared global coherency state
US20060155934A1 (en) System and method for reducing unnecessary cache operations
US20030009623A1 (en) Non-uniform memory access (NUMA) data processing system having remote memory cache incorporated within system memory
US20030009634A1 (en) Non-uniform memory access (NUMA) data processing system that provides notification of remote deallocation of shared data
US6560681B1 (en) Split sparse directory for a distributed shared memory multiprocessor system
US6292872B1 (en) Cache coherency protocol having hovering (H) and recent (R) states
US6397303B1 (en) Data processing system, cache, and method of cache management including an O state for memory-consistent cache lines
KR101072174B1 (en) System and method for implementing an enhanced hover state with active prefetches
US7024520B2 (en) System and method enabling efficient cache line reuse in a computer system
US6275908B1 (en) Cache coherency protocol including an HR state
US6687795B2 (en) Data processing system and method of communication that reduce latency of write transactions subject to retry
US6356982B1 (en) Dynamic mechanism to upgrade o state memory-consistent cache lines
US6442653B1 (en) Data processing system, cache, and method that utilize a coherency state to indicate the latency of cached data
US6349368B1 (en) High performance mechanism to support O state horizontal cache-to-cache transfers
US6415358B1 (en) Cache coherency protocol having an imprecise hovering (H) state for instructions and data
US7213109B1 (en) System and method for providing speculative ownership of cached data based on history tracking
US6944721B2 (en) Asynchronous non-blocking snoop invalidation
US6272603B1 (en) Cache coherency protocol having hovering (H), recent (R), and tagged (T) states

Legal Events

Date Code Title Description
AS Assignment

Owner name: INTERNATIONAL BUSINESS MACHINES CORPORATION, NEW Y

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:ARIMILLI, RAVI K.;ARIMILLI, LAKSHMINARAYANA B.;FIELDS, JAMES S., JR.;AND OTHERS;REEL/FRAME:010066/0870

Effective date: 19990623

FPAY Fee payment

Year of fee payment: 4

FPAY Fee payment

Year of fee payment: 8

REMI Maintenance fee reminder mailed
LAPS Lapse for failure to pay maintenance fees
STCH Information on status: patent discontinuation

Free format text: PATENT EXPIRED DUE TO NONPAYMENT OF MAINTENANCE FEES UNDER 37 CFR 1.362

FP Lapsed due to failure to pay maintenance fee

Effective date: 20140528